Power utilization scheduling and power utilization protection method based on server calculation

Abstract

The invention belongs to the technical field of server power consumption management, and discloses a power consumption scheduling and power consumption protection method based on server calculation, which comprises the following steps: step 1, data acquisition and preprocessing, step 2, construction of a power consumption prediction algorithm library, step 3, construction and solution of a power consumption prediction model, step 4, calculation of a power consumption prediction algorithm library, and step 5, calculation of a power consumption prediction model. 5, generating, regulating and controlling an optimal power consumption prediction result; and 6, capping and protecting power consumption. According to the method, the power consumption prediction algorithm library is meticulously constructed, high-precision prediction of the power consumption of the server can be realized, so that the prediction accuracy and reliability are greatly improved, and on the basis of the accurate prediction results, scheduling and optimization of power resources can be intelligently carried out, so that the power consumption of the server is greatly reduced. Therefore, the most reasonable distribution of the power in each link of the data center is ensured, the power supply and stability of the server are ensured, the overall energy consumption is effectively reduced, and the operation efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of server power management, and specifically relates to a power scheduling and power protection method based on server computing. Background Art

[0002] In the current information society, as the core equipment for data processing and storage, the energy consumption problem of servers is becoming increasingly prominent. With the rapid development of technologies such as cloud computing and big data, the scale of server clusters continues to expand, and energy consumption has also increased dramatically. This has not only led to an increase in operating costs, but also brought huge pressure on environmental protection. Traditional power scheduling methods are often based on simple statistics and analysis of historical power consumption data, which makes it difficult to accurately predict the future power demand of servers, resulting in inflexible power grid scheduling and low energy utilization efficiency. At the same time, when the server is running under high load, the power consumption will increase sharply. If there is a lack of effective protection measures, it is easy to cause overload damage or even paralysis of the entire system. Therefore, there is an urgent need for a power scheduling and power protection method based on server computing. This method can collect the power consumption data and related parameters of the server in real time, accurately predict the power consumption of the server by building a power consumption prediction algorithm library, and intelligently schedule the power grid according to the prediction results. At the same time, the power consumption cap value is set to ensure that the server works in a safe and efficient operating state, reduce energy consumption, improve energy utilization efficiency, and provide strong support for the sustainable development of green computing and data centers. Summary of the invention

[0003] The purpose of the present invention is to provide a power scheduling and power protection method based on server computing to solve the problems raised in the above background technology.

[0004] In order to achieve the above object, the present invention provides the following technical solution: a method for power dispatching and power protection based on server computing, the method comprising the following steps:

[0005] Step 1: Data collection and preprocessing: Collect the power consumption data of the server in real time, clean and preprocess the collected data, remove outliers and noise data, and ensure data quality;

[0006] Step 2: Build a power consumption prediction algorithm library:

[0007] Establish an algorithm library containing multiple electricity consumption prediction algorithms, configure key parameters in the algorithm library, and record the version number of each configuration for tracking and management to ensure algorithm performance;

[0008] Step 3: Construction and solution of power consumption prediction model:

[0009] (1) Use the algorithm and version number to solve the power consumption prediction of the preprocessed input data and obtain the power consumption prediction result for the prediction period;

[0010] (2) Compare the electricity consumption forecast results for the forecast period with the historical electricity consumption and calculate the forecast deviation.

[0011] (3) Visualize the difference between the predicted results and the actual values ​​through charts, so as to locate the cause of the prediction deviation;

[0012] Step 4: Review analysis and algorithm optimization:

[0013] Generate a re-analysis report, record the forecast deviation in detail, adjust the key parameters of the algorithm in the algorithm library based on the cause of the forecast deviation, use the adjusted algorithm to solve the input data again, obtain better power consumption forecast results, and improve forecast accuracy;

[0014] Step 5: Generate and regulate the optimal electricity consumption forecast results:

[0015] Based on the optimal power consumption forecast results, the formula is used to calculate the control instructions of the equipment on each side of the power grid. According to the control instructions, the equipment in the power grid is intelligently dispatched, including adjusting the generator output, optimizing the transmission line load, adjusting the transformer tap position, and controlling the distribution switch to ensure the safe, stable operation and efficient utilization of the power grid;

[0016] Step 6: Power consumption capping and protection:

[0017] Set a server power consumption cap as the upper limit constraint for power scheduling. When it is predicted that the server power consumption is about to exceed the cap, the protection mechanism is automatically triggered to control the server power consumption below the cap by adjusting the CPU frequency and shutting down unnecessary service measures, thus avoiding overload damage and energy waste.

[0018] Preferably, the prediction deviation calculation formula for constructing and solving the power consumption prediction model in step 3 is:

[0019]

[0020] Among them, $Q_{\text{prediction},i}$ is the predicted power consumption at the $i$th time point in the prediction period, $Q_{\text{actual},i}$ is the actual power consumption at the corresponding time point, and $k$ is the number of time points in the prediction period.

[0021] Preferably, the control instructions for generating and controlling the optimal power consumption forecast result in step 5 are:

[0022] Control command = f(Q 预测 ,Q 封顶 , device status)

[0023] Among them, $Q_{\text{forecast}}$ is the optimal power consumption forecast result, $Q_{\text{cap}}$ is the set power consumption capping value, and $\text{device status}$ is the current status of the equipment on the power generation side, transmission side, substation side, and distribution side in the power grid.

[0024] Preferably, the algorithm types of the algorithm library for constructing the electricity consumption prediction algorithm library in step 2 include but are not limited to long short-term memory network LSTM, support vector regression SVR, and random forest RF.

[0025] Preferably, the power consumption prediction solution formula for constructing and solving the power consumption prediction model in step 3 is:

[0026] Q pred =f(X,θ)

[0027] Among them, Q pred represents the electricity consumption forecast result of the forecast period, X represents the input data, θ represents the algorithm parameters, and f represents the forecast algorithm.

[0028] Preferably, the power dispatching and power protection system based on the above method includes: a data acquisition module, a data preprocessing module, a prediction algorithm library module, a power consumption prediction module, a replay analysis and algorithm optimization module, and a power dispatching and protection module.

[0029] Preferably, the data acquisition module comprises a hardware component power consumption data acquisition unit and a server environmental parameter data acquisition unit, which are respectively used to collect the power consumption data of the server hardware components and the utilization rate and environmental temperature parameters of the server in real time.

[0030] Preferably, the prediction algorithm library module includes an algorithm storage unit, a parameter configuration unit and a version management unit;

[0031] The algorithm storage unit is used to store a variety of electricity consumption prediction algorithms, the parameter configuration unit is used to configure the key parameters of the algorithm, and the version management unit is used to record the version number of each configuration in order to track and manage the algorithm performance.

[0032] Preferably, the power consumption prediction module includes a prediction solution unit and a prediction deviation calculation unit. The prediction solution unit is used to use an algorithm and a version number to solve the power consumption prediction for the preprocessed input data; the prediction deviation calculation unit is used to compare the power consumption prediction result of the prediction period with the historical power consumption, calculate the prediction deviation, and visualize the difference between the prediction result and the actual value through a chart.

[0033] Preferably, the power scheduling and protection module includes a control instruction calculation unit and an intelligent scheduling unit; the control instruction calculation unit is used to calculate the control instructions of the equipment on each side of the power grid based on the optimal power consumption prediction result, and the intelligent scheduling unit is used to intelligently schedule the equipment in the power grid according to the control instructions and set the server power consumption capping value.

[0034] The beneficial effects of the present invention are as follows:

[0035] By carefully constructing a power consumption prediction algorithm library, the present invention can achieve high-precision prediction of server power consumption, thereby greatly improving the accuracy and reliability of the prediction. This precise prediction lays a solid foundation for subsequent power management. Secondly, based on these accurate prediction results, the present invention can intelligently dispatch and optimize power resources to ensure that power is most reasonably distributed in various links of the data center, effectively reducing overall energy consumption and improving operational efficiency. In addition, the present invention also specifically sets a power consumption cap value, and can automatically trigger a protection mechanism when the threshold is reached. This design effectively avoids the risk of damage to the server due to overload, and provides a strong guarantee for the safe and stable operation of the power grid. Finally, by continuously optimizing algorithms and intelligent scheduling strategies, the present invention greatly improves the efficiency of energy utilization, not only reduces operating costs, but also actively responds to the call for green development, and injects strong impetus into the sustainable development of data centers. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A simplified flow chart of the method steps for power dispatching and protection of the present invention;

[0037] Figure 2 This is a composition diagram of the power dispatching and protection system of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] like Figure 1 to Figure 2 As shown, an embodiment of the present invention provides a method for power scheduling and power protection based on server computing, the method comprising the following steps:

[0040] Step 1: Data collection and preprocessing: Real-time collection of server power consumption data, including power consumption data of hardware components such as CPU, memory, and hard disk, as well as server utilization and ambient temperature parameters. Clean and preprocess the collected data to remove outliers and noise data to ensure data quality.

[0041] Step 2: Build a power consumption prediction algorithm library:

[0042] Establish an algorithm library containing multiple electricity consumption prediction algorithms, configure key parameters in the algorithm library, such as learning rate, number of iterations, regularization parameters, etc., and record the version number of each configuration for tracking and management to ensure algorithm performance;

[0043] Step 3: Construction and solution of power consumption prediction model:

[0044] (1) Use the algorithm and version number to solve the power consumption prediction of the preprocessed input data and obtain the power consumption prediction result for the prediction period;

[0045] (2) Compare the electricity consumption forecast results for the forecast period with the historical electricity consumption and calculate the forecast deviation.

[0046] (3) Visualize the difference between the predicted results and the actual values ​​through charts, so as to locate the cause of the prediction deviation;

[0047] Step 4: Review analysis and algorithm optimization:

[0048] Generate a replay analysis report, and record the forecast deviation in detail, including the size of the deviation, the time of occurrence, possible causes, etc. Based on the cause of the forecast deviation, adjust the key parameters of the algorithm in the algorithm library, such as adjusting the learning rate, increasing the feature dimension, optimizing the model structure, etc. Use the adjusted algorithm to solve the input data again to obtain a better power consumption forecast result and improve the forecast accuracy;

[0049] Step 5: Generate and regulate the optimal electricity consumption forecast results:

[0050] Based on the optimal power consumption forecast results, the formula is used to calculate the control instructions of the equipment on each side of the power grid. According to the control instructions, the equipment in the power grid is intelligently dispatched, including adjusting the generator output, optimizing the transmission line load, adjusting the transformer tap position, and controlling the distribution switch to ensure the safe, stable operation and efficient utilization of the power grid;

[0051] Step 6: Power consumption capping and protection:

[0052] Set a server power consumption cap as the upper limit constraint for power scheduling. When it is predicted that the server power consumption is about to exceed the cap, the protection mechanism is automatically triggered to control the server power consumption below the cap by adjusting the CPU frequency and shutting down unnecessary service measures, thus avoiding overload damage and energy waste.

[0053] Among them, the prediction deviation calculation formula for building and solving the power consumption prediction model in step 3 is:

[0054]

[0055] Among them, $Q_{\text{prediction},i}$ is the predicted power consumption at the $i$th time point in the prediction period, $Q_{\text{actual},i}$ is the actual power consumption at the corresponding time point, and $k$ is the number of time points in the prediction period.

[0056] Among them, the control instructions for generating and controlling the optimal power consumption forecast results in step 5 are:

[0057] Control command = f(Q 预测 ,Q 封顶 , device status)

[0058] Among them, $Q_{\text{forecast}}$ is the optimal power consumption forecast result, $Q_{\text{cap}}$ is the set power consumption capping value, and $\text{device status}$ is the current status of the equipment on the power generation side, transmission side, substation side, and distribution side in the power grid.

[0059] Among them, the algorithm types of the algorithm library for constructing the electricity consumption prediction algorithm library in step 2 include but are not limited to long short-term memory network LSTM, support vector regression SVR, and random forest RF.

[0060] Among them, the power consumption prediction solution formula for constructing and solving the power consumption prediction model in step 3 is:

[0061] Q pred =f(X,θ)

[0062] Among them, Q pred represents the electricity consumption forecast result of the forecast period, X represents the input data, θ represents the algorithm parameters, and f represents the forecast algorithm.

[0063] Among them, the power dispatching and power protection system based on the above method includes: data acquisition module, data preprocessing module, prediction algorithm library module, power consumption prediction module, review analysis and algorithm optimization module, power dispatching and protection module.

[0064] The data acquisition module includes a hardware component power consumption data acquisition unit and a server environmental parameter data acquisition unit, which are respectively used to collect the power consumption data of the server hardware components and the utilization rate and environmental temperature parameters of the server in real time.

[0065] Among them, the prediction algorithm library module includes an algorithm storage unit, a parameter configuration unit and a version management unit;

[0066] The algorithm storage unit is used to store a variety of electricity consumption prediction algorithms, the parameter configuration unit is used to configure the key parameters of the algorithm, and the version management unit is used to record the version number of each configuration in order to track and manage the algorithm performance.

[0067] Among them, the power consumption prediction module includes a prediction solution unit and a prediction deviation calculation unit. The prediction solution unit is used to solve the power consumption prediction of the preprocessed input data using an algorithm and a version number; the prediction deviation calculation unit is used to compare the power consumption prediction results of the prediction period with the historical power consumption, calculate the prediction deviation, and visualize the difference between the prediction results and the actual values ​​through charts.

[0068] Among them, the power dispatching and protection module includes a control instruction calculation unit and an intelligent dispatching unit; the control instruction calculation unit is used to calculate the control instructions of the equipment on each side of the power grid based on the optimal power consumption forecast results, and the intelligent dispatching unit is used to intelligently dispatch the equipment in the power grid according to the control instructions and set the server power consumption cap value.

[0069] Embodiment 1

[0070] This embodiment provides a power consumption scheduling and protection system based on server computing power prediction, including a data acquisition module, a preprocessing module, a power consumption prediction algorithm library, a power consumption prediction module, a replay analysis module, a control module and a protection module. Each module works together to achieve accurate prediction of server power consumption and intelligent scheduling and protection.

[0071] Embodiment 2

[0072] Taking a large data center as an example, the method and system of the present invention are used for power scheduling and protection. First, the server power consumption data is collected in real time through the data acquisition module, and cleaned and preprocessed by the preprocessing module. Then, the LSTM algorithm in the power consumption prediction algorithm library is used to solve the preprocessed data to obtain the power consumption prediction results for the prediction period. The prediction results are compared with the historical power consumption, and it is found that there is a prediction deviation. The re-display analysis report is generated by the re-display analysis module, and the key parameters of the LSTM algorithm are adjusted based on the deviation reasons in the report. After adjustment, the prediction is performed again to generate the optimal power consumption prediction result. Finally, according to the optimal prediction result, the state of relevant equipment in the power grid is regulated by the control module to ensure the safe and stable operation of the power grid. At the same time, the server power consumption capping value is set. When it is predicted that the power consumption is about to exceed the standard, the protection mechanism is automatically triggered by the protection module to avoid overload damage and energy waste.

[0073] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0074] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for power dispatching and power protection based on server computing, characterized in that: The method comprises the following steps: Step 1: Data collection and preprocessing: Collect the power consumption data of the server in real time, clean and preprocess the collected data, remove outliers and noise data, and ensure data quality; Step 2: Build a power consumption prediction algorithm library: Establish an algorithm library containing multiple electricity consumption prediction algorithms, configure key parameters in the algorithm library, and record the version number of each configuration for tracking and management to ensure algorithm performance; Step 3: Construction and solution of power consumption prediction model: (1) Use the algorithm and version number to solve the power consumption prediction of the preprocessed input data and obtain the power consumption prediction result for the prediction period; (2) Compare the electricity consumption forecast results for the forecast period with the historical electricity consumption and calculate the forecast deviation; (3) Visualize the difference between the predicted results and the actual values ​​through charts, so as to locate the cause of the prediction deviation; Step 4: Review analysis and algorithm optimization: Generate a re-analysis report, record the forecast deviation in detail, adjust the key parameters of the algorithm in the algorithm library based on the cause of the forecast deviation, use the adjusted algorithm to solve the input data again, obtain better power consumption forecast results, and improve forecast accuracy; Step 5: Generate and regulate the optimal electricity consumption forecast results: Based on the optimal power consumption forecast results, the formula is used to calculate the control instructions of the equipment on each side of the power grid. According to the control instructions, the equipment in the power grid is intelligently dispatched, including adjusting the generator output, optimizing the transmission line load, adjusting the transformer tap position, and controlling the distribution switch to ensure the safe, stable operation and efficient utilization of the power grid; Step 6: Power consumption capping and protection: Set a server power consumption cap as the upper limit constraint for power scheduling. When it is predicted that the server power consumption is about to exceed the cap, the protection mechanism is automatically triggered to control the server power consumption below the cap by adjusting the CPU frequency and shutting down unnecessary service measures, thus avoiding overload damage and energy waste.

2. The method for power dispatching and power protection based on server computing according to claim 1, characterized in that: The prediction deviation calculation formula for constructing and solving the power consumption prediction model in step 3 is: Among them, $Q_{\text{prediction},i}$ is the predicted power consumption at the $i$th time point in the prediction period, $Q_{\text{actual},i}$ is the actual power consumption at the corresponding time point, and $k$ is the number of time points in the prediction period.

3. The method for power dispatching and power protection based on server computing according to claim 1, characterized in that: The control instructions for generating and controlling the optimal power consumption forecast results in step 5 are: Control command = f(Q 预测 ,Q 封顶 , device status) Among them, $Q_{\text{forecast}}$ is the optimal power consumption forecast result, $Q_{\text{cap}}$ is the set power consumption capping value, and $\text{device status}$ is the current status of the equipment on the power generation side, transmission side, substation side, and distribution side in the power grid.

4. The method for power dispatching and power protection based on server computing according to claim 1, characterized in that: The algorithm types of the algorithm library for constructing the electricity consumption prediction algorithm library in step 2 include but are not limited to long short-term memory network LSTM, support vector regression SVR, and random forest RF.

5. The method for power dispatching and power protection based on server computing according to claim 1, characterized in that: The power consumption prediction formula for constructing and solving the power consumption prediction model in step 3 is: Q pred =f(X,θ) Among them, Q pred represents the electricity consumption forecast result of the forecast period, X represents the input data, θ represents the algorithm parameters, and f represents the forecast algorithm.

6. A method for power dispatching and power protection based on server computing according to claims 1-5, characterized in that: The power dispatching and protection system based on the above method includes: data acquisition module, data preprocessing module, prediction algorithm library module, power consumption prediction module, replay analysis and algorithm optimization module, power dispatching and protection module.

7. The method for power dispatching and power protection based on server computing according to claim 6 is characterized in that: The data acquisition module includes a hardware component power consumption data acquisition unit and a server environmental parameter data acquisition unit, which are respectively used to collect the power consumption data of the server hardware components and the utilization rate and environmental temperature parameters of the server in real time.

8. The method for power dispatching and power protection based on server computing according to claim 6 is characterized in that: The prediction algorithm library module includes an algorithm storage unit, a parameter configuration unit and a version management unit; The algorithm storage unit is used to store a variety of electricity consumption prediction algorithms, the parameter configuration unit is used to configure the key parameters of the algorithm, and the version management unit is used to record the version number of each configuration in order to track and manage the algorithm performance.

9. The method for power dispatching and power protection based on server computing according to claim 6, characterized in that: The power consumption prediction module includes a prediction solution unit and a prediction deviation calculation unit. The prediction solution unit is used to solve the power consumption prediction for the preprocessed input data using an algorithm and a version number; the prediction deviation calculation unit is used to compare the power consumption prediction result of the prediction period with the historical power consumption, calculate the prediction deviation, and visualize the difference between the prediction result and the actual value through a chart.

10. The method for power dispatching and power protection based on server computing according to claim 6, characterized in that: The power dispatching and protection module includes a control instruction calculation unit and an intelligent dispatching unit; The control instruction calculation unit is used to calculate the control instructions of the equipment on each side of the power grid based on the optimal power consumption prediction result, and the intelligent scheduling unit is used to intelligently schedule the equipment in the power grid according to the control instructions and set the server power consumption cap value.

Images